Rule-List Fallacy

Matthew Dryer dryer at
Thu Jun 19 01:21:54 UTC 2008

I think Tom's quite right about the issue of what is the default being 
something of a red herring.  I confess to somewhat opportunistically using 
that topic as a segue into a related but more substantive point, which is 
that there is a widespead tendency of linguists of many stripes - not just 
generative linguists, but construction grammarians, cognitive semanticists, 
and even some deeply involved in grammaticalization - to assume that where 
we linguists find patterns in language, speakers must find the same 
patterns as well.

I should also point out, in response to Edith's comment, that the case I 
was commenting on is actually not one that involves "people holistically 
memorizing expressions that are usually analyzed as adpositional phrases". 
Under the alternative I was proposing, people do analyse adpositional 
phrases into "adposition" plus noun phrase; it's just that people don't 
recognize that different "adpositions" all do the same thing.  I.e. they 
don't recognize the pattern and thus don't reocgnize a class of adpositions.

Unfortunately, I'm guilty of what Fritz was originally accused of.  I've 
given a (brief) abstract of a paper without providing the arguments that 
are in the paper (which I haven't written up yet).


--On Wednesday, June 18, 2008 1:30 PM -0700 Tom Givon <tgivon at> 

> It seems to me that this discussion gravitates, slowly but surely,
> towards an empirical vacuum. Perhaps a better learning strategy for all
> of us would be to concede that, in principle, both holistic and analyzed
> processing of complex expressions are in principle ('cognitively')
> available to speakers. Then find methods by which we can tell one
> strategy (as used by speakers/hearers) from the other; then find contexts
> that induce speakers/hearers to prefer one strategy over the other; and
> then--voila, denouement--count the frequency distribution of the two
> types of behavior in various contexts. To simply express personal
> preference for one over the other as 'the default strategy' is not all
> that helpful. Cheers,  TG
> ========
> David Tuggy wrote:
>> Edith Moravcsik wrote:
>>> <snip>If we did not know that these expressions might be analyzed by
>>> linguists as multi-part phrases, there would be nothing suprising
>>> about how people treat them; and we would lose the interesting
>>> question of why linguists' analyses and people's ways of processing
>>> these expressions parted ways.
>>> The same holds for formulaic expressions in general. The reason it is
>>> interesting that people treat them as atomic wholes is that we
>>> linguists can analyze them as having parts.
>> Well, it is also interesting that people can also analyze them as
>> having parts. Linguists are people too, of course, but non-linguist
>> people are often quite aware of parts of formulaic structures. The
>> fact that both modes are available (though perhaps differentially
>> attractive) to both linguists and language speakers is, I would
>> maintain, highly important (as well as interesting).
>> --David Tuuggy

More information about the Funknet mailing list